01
The change
Claude, GPT, and their peers can already produce a competent financial model and answer sophisticated analytical questions. Access is getting cheaper, broader, and less differentiated.
We build the context layer between domain expertise and frontier intelligence.
IRIS is our product forfinancial models and fundamental investment research,yielding higher analytical resolution per inference dollar.
"The IRIS Model"
Import a financial model IRIS has never seen. IRIS figures out how it works and builds a Method for each important line. The Method runs inside the spreadsheet through an =AI() formula. The analyst teaches IRIS where the investment view differs by editing the Method in plain English.
The payoff is the IRIS model: the analyst's differentiated view, with every change on the record.
The premise
Model building is getting easier. Edge is still hard.
Frontier models are becoming ordinary tools for analysts. The hard part is no longer getting access to intelligence. It is giving that intelligence enough context to do serious work.
02
The constraint
But every serious question starts by rebuilding context: What are we analyzing? What does the number mean? Which assumptions, Method, evidence, and prior call matter?
03
The durable unit of work
Fundamental research already has a durable home for much of that context: the financial model. History becomes estimates; estimates become valuation and outputs.
04
The gap
The problem is that the spreadsheet stores the calculation better than the meaning. An analyst sees same-store sales margin expansion under current guidance. The grid sees =H32/H18.
05
IRIS
IRIS supplies the missing meaning. It reads the unfamiliar workbook and builds a Method for each important line: a plain-English statement of how that line works, run inside the spreadsheet through =AI().
The analyst edits the pre-built Method to teach IRIS where the investment view differs, runs the model, and sees the consequence in the numbers. The baseline becomes the IRIS model: the investor's differentiated view.
The learning flywheel
Start with a baseline. Teach it your view.
The analyst does not rebuild the workbook. IRIS turns the reasoning it infers into pre-built Methods inside the model. The analyst edits a Method in plain English, and the Run shows the consequence.
01 / IMPORT
Baseline model
Import a competent, consensus-like workbook from Claude, GPT, or anywhere else.
02 / INFER
Methods in the model
IRIS infers how each important line works and builds a pre-built Method. An =AI() formula runs that Method inside the spreadsheet.
03 / TEACH
Your view
The analyst teaches IRIS by editing a Method in plain English where the investment view differs.
04 / RUN
The consequence
Run the model and see the changed assumption move through the outputs and valuation.
05 / DIFF
The IRIS model
Compare versions. See what changed, why it changed, and what the judgment affected.
The IRIS model is the investor's differentiated view, captured as a governed and reusable research record.
What IRIS understands
The model is more than a grid.
Much of fundamental modeling is judgment expressed as heuristics and rubrics. The grid stores the calculation better than the reasoning behind it.
Model Logic is how the model thinks.
Each analytical responsibility has an identity, formula, Method, evidence, dependencies, and known unknowns. The Method says how the line should work in plain English. =AI() runs it inside the financial model. The analyst teaches IRIS by editing the Method, without rebuilding the workbook. That is how IRIS narrows 11,112 cells to one question without losing the model around it.
Model Memory is how that thinking evolves.
Versions, Runs, Results, and Timeline preserve the prior call, the facts behind it, what changed, and whether the judgment proved right. The full record remains intact; the relevant history comes forward.
What just works
Ask. Explain. Verify.
Once the model has learned the analyst's view, it becomes something the analyst can interrogate in plain English.
01 / Ask
Ask the question you actually have.
Select an output, assumption, or driver. Ask why it changed, what drives it, or what the analyst believed last time.
02 / Explain
Get the model in plain English.
IRIS connects the formulas to the Method, assumptions, evidence, and history behind the answer.
03 / Verify
See where the answer came from.
Trace the explanation back to the exact model, cell, Method, evidence, Version, and prior call.
No more model archaeology.
Stop tracing precedents and reconstructing someone else's logic before you can answer a basic question.
No new way of working.
Use IRIS, Excel, or Claude. The workbook stays put while IRIS carries the reasoning with the question.
Why governance matters
Every change and answer comes with a trail.
IRIS can show exactly which company, model, subject, Method, Version, evidence, and lineage produced the change or answer.
Governance is not another task for the analyst. It is built into the work.
Exact identities. Traceable evidence. Versioned judgment. A record of what changed.
The payoff
Turn the baseline into your investment view.
Teach one Method, run the model, and put the consequence on the record.